AI did not kill Google. It handed it a bigger job.

Somewhere in early 2023, the obituary was written. ChatGPT had arrived, people were asking it questions they used to type into Google, and the consensus formed fast: search was over, the ten blue links were a museum piece, and Google's ad business was about to unwind. Three years later we are still waiting, and the obituary gets quietly reissued every spring. Alphabet's search revenue grew at a double-digit rate again in 2025. AI Overviews reach, by Google's own count, more than two billion people a month. The company that was supposed to be disrupted is the largest AI search engine on earth.

We are not Google cheerleaders. Two of us spent six years inside it and left, which is usually a sign of something. But we do read the numbers, and the numbers say the story everyone told was wrong in an interesting way.

Why the prediction failed

The prediction assumed that search was a product. It is not. Search is a habit connected to a distribution machine: the default box in Chrome, on Android, in Safari, on every iPhone, and a two-decade index that knows which pages can be trusted. ChatGPT changed how a slice of questions get asked. It did not change where the majority of the world's intent shows up when someone wants to buy something.

It also assumed Google would not respond. Google shipped AI Overviews, then AI Mode, and put its own model on top of its own index. The habit stayed, the interface changed, and the ads followed the interface.

What actually changed

Three things, and they all matter to anyone spending money on search.

The click became optional. For a growing share of informational queries, the answer is on the results page and the person leaves satisfied. Traffic to "what is" and "how does" content has fallen, in some categories sharply. If your content strategy was built on those pages, you have felt it.

The passage replaced the page. Google now cites fragments. The page that ranks third but has the clearest paragraph gets quoted above the one that ranks first and buries the answer. We wrote a full guide on how to appear in AI Overviews.

The buyer arrives later and better informed. People do their research in an AI interface, then search for a brand or a product by name. Branded and high-intent queries are growing while generic informational ones shrink. Paid search on those queries has become more valuable, not less.

What did not change

Intent still converts. A person typing "best running shoes for flat feet" and clicking an ad is still the most valuable visitor most brands will get this week. Google still sells that click, and the auction is still where the most measurable money in advertising changes hands. Every client we run paid search for has seen the channel hold or grow through the "death of search" years, provided the account was run on real conversion data rather than on the platform's optimism.

Where the real risk is

Not in Google disappearing. In brands mis-reading the shift and cutting the wrong thing.

• Cutting search because "people use ChatGPT now", while their own branded queries rise every quarter.
• Keeping an SEO strategy built on 2019 informational content, watching it lose traffic, and concluding SEO is dead rather than that the content is.
• Treating GEO as a separate discipline with a separate agency, when it is the same craft applied to a slightly different reader. Our take on that: the indispensables of GEO.

Our position, in one paragraph

Google had AI running its auctions, its bidding and its query understanding for a decade before anyone called it AI. What happened in 2023 is that the interface caught up with the engine. The job for a brand is the same as it has always been, with a harder exam: be the clearest, most trustworthy source on the questions your buyers actually ask, measure what comes back on your own data, and put paid money where intent is highest. The tools changed. The discipline did not. If anything, the discipline is now the whole game.